seasonal cycle of volume transport through kerama gap revealed … · 2015. 11. 30. ·...
TRANSCRIPT
-
Ocean Modelling 96 (2015) 203–213
Contents lists available at ScienceDirect
Ocean Modelling
journal homepage: www.elsevier.com/locate/ocemod
Seasonal cycle of volume transport through Kerama Gap revealed by a
20-year global HYbrid Coordinate Ocean Model reanalysis
Zhitao Yu a,b,∗, E. Joseph Metzger b, Prasad Thoppil b, Harley E. Hurlburt c, Luis Zamudio c,Ole Martin Smedstad d, Hanna Na e, Hirohiko Nakamura f, Jae-Hun Park g
a American Society of Engineering Education, DC, USAb Naval Research Laboratory, Stennis Space Center, MS, USAc Florida State University, Tallahassee, FL, USAd Vencore, Incorporated, Stennis Space Center, MS, USAe Faculty of Science, Hokkaido University, Sapporo, Japanf Faculty of Fisheries, Kagoshima University, Kagoshima, Japang Department of Ocean Sciences, Inha University, Incheon, South Korea
a r t i c l e i n f o
Article history:
Received 3 April 2015
Revised 16 October 2015
Accepted 31 October 2015
Available online 10 November 2015
Keywords:
Kuroshio
Mesoscale eddy
HYCOM
Kerama Gap
a b s t r a c t
The temporal variability of volume transport from the North Pacific Ocean to the East China Sea (ECS) through
Kerama Gap (between Okinawa Island and Miyakojima Island − a part of Ryukyu Islands Arc) is investigatedusing a 20-year global HYbrid Coordinate Ocean Model (HYCOM) reanalysis with the Navy Coupled Ocean
Data Assimilation from 1993 to 2012. The HYCOM mean transport is 2.1 Sv (positive into the ECS, 1 Sv =106 m3/s) from June 2009 to June 2011, in good agreement with the observed 2.0 Sv transport during the
same period. This is similar to the 20-year mean Kerama Gap transport of 1.95 ± 4.0 Sv. The 20-year monthlymean volume transport (transport seasonal cycle) is maximum in October (3.0 Sv) and minimum in Novem-
ber (0.5 Sv). The annual variation component (345–400 days), mesoscale eddy component (70–345 days),
and Kuroshio meander component (< 70 days) are separated to determine their contributions to the trans-
port seasonal cycle. The annual variation component has a close relation with the local wind field and in-
creases (decreases) transport into the ECS through Kerama Gap in summer (winter). Most of the variations
in the transport seasonal cycle come from the mesoscale eddy component. The impinging mesoscale eddies
increase the transport into the ECS during January, February, May, and October, and decrease it in March,
April, November, and December, but have little effect in summer (June–September). The Kuroshio meander
components cause smaller transport variations in summer than in winter.
© 2015 Elsevier Ltd. All rights reserved.
1
r
n
t
(
w
T
E
e
s
t
S
h
p
(
t
n
S
m
t
a
j
o
h
1
. Introduction
The Kuroshio is one of the world’s major western boundary cur-
ents and a key feature of North Pacific Ocean circulation. It origi-
ates from the North Equatorial Current (Gordon et al., 2014) and
hen enters the East China Sea (ECS) through the East Taiwan Channel
ETC) between Taiwan and Ishigaki Island; it carries warm and saline
aters poleward (Oka and Kawabe, 1998), and exits the ECS through
okara Strait (Fig. 1). Estimates of the mean Kuroshio transport in the
CS vary from 18.5 to 32 Sv (Roemmich and McCallister, 1989; Johns
t al., 2001; Andres et al., 2008b). Because the Kuroshio transports
ignificant amounts of heat, salt, and mass from the tropical ocean
o mid-latitudes, it has a great influence on the global climate and
∗ Corresponding author at: Oceanography Division, Naval Research Laboratory,tennis Space Center, MS 39529, USA. Tel.: +2286884883.
E-mail address: [email protected] (Z. Yu).
fl
b
(
t
(
ttp://dx.doi.org/10.1016/j.ocemod.2015.10.012
463-5003/© 2015 Elsevier Ltd. All rights reserved.
eat balances (Qu and Lukas, 2003), and on the fisheries, hydrogra-
hy, and weather of countries surrounding the Northwestern Pacific
Qiu, 2001).
The Ryukyu Islands Arc forms a barrier along the eastern side of
he ECS and separates the ECS from the North Pacific except at con-
ecting straits. Thus, except at the entrance (ETC) and exit (Tokara
trait), the water in the ECS and the North Pacific exchanges through
any channels in the Ryukyu Islands Arc. Kerama Gap, located be-
ween Miyakojima and Okinawa (Fig. 1), is the deepest channel with
sill depth of 1050 m (Choi et al., 2002) and thus has been the sub-
ect of significant research. Analyzing moored current meter (CM)
bservations, Yuan et al. (1994) reported an observed 5.8 Sv out-
ow (from the ECS to the Northwestern Pacific) through the passages
etween Miyakojima and Okinawa during Fall 1991; but Yuan et al.
1995) estimated a 2.4 Sv inflow (from the Northwestern Pacific into
he ECS) from November 1991 to September 1992. Morinaga et al.
1998) estimated a 7.2 Sv inflow through Kerama Gap from their CM
http://dx.doi.org/10.1016/j.ocemod.2015.10.012http://www.ScienceDirect.comhttp://www.elsevier.com/locate/ocemodhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.ocemod.2015.10.012&domain=pdfmailto:[email protected]://dx.doi.org/10.1016/j.ocemod.2015.10.012
-
204 Z. Yu et al. / Ocean Modelling 96 (2015) 203–213
Longitude (oE)
Latit
ude
(oN
)
5
4 ETC 3
2
1
Toka
ra S
trait
122 124 126 128 130 13223
24
25
26
27
28
29
30
31
32
Dep
th (
m)
0
100
200
300
400
500
600
700
800
900
1000
2000
3000
4000
5000
6000
Fig. 1. HYCOM bathymetry (meters) for the East China Sea. Gray represents model
land points. Okinawa (1), Miyakojima (2), Ishigaki (3), the East Taiwan Channel (ETC),
Taiwan (4), Kyushu (5), and Tokara Strait are labeled. The black line represents the PN-
line. The land -ocean boundary in HYCOM is defined by the 10 cm isobath but all depths
less than 5 m are set to 5 m. A zoom centered on Kerama Gap is inset in the lower right
corner. The gray solid line represents the HYCOM transect used to determine transport
through Kerama Gap. The four red dots represent the locations of CPIES moorings ES1
to ES4 and three white dots represent the locations of current meters CM1 to CM3
(SW-NE, Na et al., 2014). The green dot represents the location of CPIES mooring ES5.
(For interpretation of the references to color in this figure legend, the reader is referred
to the web version of this article).
t
2
b
s
2
w
m
i
e
h
r
t
fl
o
t
a
l
s
2
a
a
f
C
o
v
o
e
a
O
t
d
a
c
t
v
s
a
2
a
d
H
a
l
t
t
b
s
s
t
t
n
c
t
W
c
c
O
t
C
m
observations during two months from July to September 1992. The
wide range of observed transports can be attributed to the relatively
short observation duration periods covering different observation
times and is indicative of the large temporal variations in transport
through Kerama Gap. The annual mean transport through Kerama
Gap remained uncertain until Na et al. (2014) reported a 2.0 Sv mean
flow into the ECS based on two years of observations covering June
2009–June 2011. The standard deviation is 3.2 Sv, which is much
larger than the 2-year mean inflow and comparable to the standard
deviation of the downstream PN-line (Fig. 1, black line) Kuroshio
transport (4.0 Sv) (Andres et al., 2008b). Hence, Kerama Gap trans-
port may have a significant impact on the temporal variability of the
Kuroshio transport in the ECS.
Classical theories have shown that mid-latitude eastward flows
have a spontaneous tendency to develop wavelike disturbances due
to baroclinic instability (Kundu and Cohen, 2002). Since the Kuroshio
Extension and the Subtropical Countercurrent both exist at mid-
latitudes, there is no surprise that mesoscale eddies are ubiquitous
outside of the ECS with the Ryukyu Islands Arc acting as a bar-
rier. Andres et al. (2008a, b) show that the eddies arriving from the
ocean interior affect the transport through Kerama Gap and then the
Kuroshio transport in the ECS. Andres and Cenedese (2013) further
found laboratory support for Andres et al. (2008a, b). Jin et al. (2010),
on the other hand, argued that a shifting of the Kuroshio mean cur-
rent axis and the approaching eddies are both important in deter-
mining flow direction through Kerama Gap. The preceding results in-
dicate the important role of the Kerama Gap in the interaction of the
ECS-Kuroshio with the ocean interior and on water exchange through
the Ryukyu Islands Arc.
In a numerical attempt to simulate Ryukyu currents with climato-
logical forcing, You and Yoon (2004) reported a 5.6 Sv inflow through
the passages between Miyakojima and Okinawa. In their model, spa-
tial resolution is 1/6°. Guo et al. (2006) ran a 1/18° nested oceanmodel using weekly forcing over the period from September 1991
to December 1998 and found an inflow of 0.49 Sv between Miyako-
jima and Okinawa. Soeyanto et al. (2014) estimated a 0.18 Sv inflow
hrough Kerama Gap by analyzing the results from a 20-year (1993–
012) reanalysis output by a data-assimilative ocean model, which is
ased on the Princeton Ocean Model with a generalized coordinate
ystem, developed in Japan Coastal Ocean Predictability Experiments
. Their model includes two sub-models that are connected by a one-
ay nesting system and the horizontal grid interval for the inner
odel (10.5–60°N, 108–180°E) is 1/12°. All of these numerical stud-es reported inflow through Kerama Gap, which is consistent with Na
t al. (2014), yet the mean transport is quite different. Kerama Gap
as very steep topography and its width is only about 50 km. Thus,
esolving the transport requires fine horizontal resolution and a ver-
ical coordinate system capable of resolving the vertical structure of
ow from the surface to the sill depth. Additionally, data assimilation
f sea surface height (SSH) is necessary for the model to capture the
emporal transport variation generated by approaching eddies.
Na et al. (2014) have investigated the dynamics at Kerama Gap
nd reported the mean transport. To elucidate the dynamics under-
ying the variation of transport through Kerama Gap, it is neces-
ary to estimate its variability at various timescales. However, the
-year observational period is not long enough to determine vari-
bility in timescales longer than one year. In this study, we present
20-year (1993–2012) transport time series through Kerama Gap
rom a data assimilative global HYbrid Coordinate Ocean Model (HY-
OM) reanalysis. The long transport time series provides a unique
pportunity that allows us to define the seasonal cycle and to in-
estigate the impact of transport variability at different time scales
n the seasonal cycle. Previous studies (Sugimoto et al., 1988; Qiu
t al., 1990; James et al., 1999; Ichikawa, 2001; Nakamura et al., 2003)
lso show that two types of Kuroshio meanders exist in the northern
kinawa trough with periods less than 70 days. Thus, we focus on
hree bands with periods of 345–400 days (annual variation), 70–345
ays (mesoscale eddy), and shorter than 70 days (Kuroshio meander)
nd also examine their respective contributions to the seasonal cy-
le. The paper is organized as follows: the numerical model used in
his study is described in Section 2. Model comparisons with obser-
ational data are presented in Section 3. Transport variability is de-
cribed in Section 4. Dynamics underlying the transport variability
re discussed in Section 5, followed by conclusions in Section 6.
. Numerical model
HYCOM is a primitive equation general ocean circulation model
pplied to large scale, marginal sea, and coastal studies. A detailed
escription of HYCOM physics can be found in Bleck (2002). Below,
YCOM is briefly presented with emphasis on the model aspects that
re relevant for this study.
HYCOM solves five prognostic equations: two for horizontal ve-
ocity components, a mass continuity equation, and two conserva-
ive equations that govern temperature and salinity. The prognos-
ic equations are time-integrated using a split-explicit treatment of
arotropic and baroclinic modes. There are three vertical-coordinate
ystems coexisting in HYCOM: z-coordinates in unstratified water,
igma-coordinates in shallow depths, and isopycnal coordinates in
he stratified ocean. Hence, HYCOM maintains the significant advan-
ages of an isopycnal model in the stratified ocean, but allows coordi-
ate surfaces to locally deviate from isopycnals to provide more verti-
al resolution near the surface and in shallow coastal regions in order
o better represent the upper ocean physics (Chassignet et al., 2003).
ith this unique feature, HYCOM serves as a good tool for simulating
irculations near Kerama Gap, which has complex topography that
overs the shallow water near Kerama Gap and Okinawa Island, the
kinawa trough, slope, and the deep ocean.
The data assimilation scheme employed for the reanalysis is a
hree dimensional variational scheme (3DVAR) used within the Navy
oupled Ocean Data Assimilation (NCODA) (Cummings, 2005; Cum-
ings and Smedstad, 2013). The ocean data sets assimilated by
-
Z. Yu et al. / Ocean Modelling 96 (2015) 203–213 205
N
a
v
N
s
t
t
A
v
e
t
r
r
n
t
s
a
p
(
a
e
s
j
l
2
3
s
t
p
i
i
t
n
a
o
t
3
s
w
c
m
t
t
T
l
v
i
p
C
d
s
fi
f
f
K
t
l
Latit
ude
(oN
)
5 cm/s
Year1 (June 2009−June 2010)
365 − 470 m
25
25.5
26
26.5
5 cm/s
Year2 (June 2010−June 2011)
443 − 548 m
Latit
ude
(oN
)
5 cm/s
620 − 725 m
25
25.5
26
26.5
5 cm/s
698 − 753 m
Latit
ude
(oN
)
Longitude (oE)
5 cm/s
975 m
126.5 127 127.5 12825
25.5
26
26.5
Longitude (oE)
5 cm/s
1008 m
126.5 127 127.5 128
a
c
e
b
d
f
Fig. 2. Mean current vectors at upper (a, b), middle (c, d), and near sill depth (e, f)
layers in Kerama Gap for year-1 (left) and year-2 (right) at CM1 (southwest) to CM3
(northeast). Only CM2 has measurements near the sill depth. Observations from Na
et al. (2014) are shown in black and the HYCOM reanalysis in red. Gray represents
model land points. The reference vector is shown in the lower right corner of each
panel. (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article).
2
1
(
t
t
t
h
s
t
t
r
d
C
t
o
t
C
fl
w
t
y
t
r
t
t
t
t
CODA include: remotely sensed sea surface temperature (SST), SSH,
nd sea ice concentration; plus in-situ surface and subsurface obser-
ations of temperature and salinity. An important component within
CODA is forming 3D synthetic profiles from the 2D SSH and SST,
ince there are only very limited subsurface profile data to constrain
he system. In the global HYCOM reanalysis, HYCOM assimilates syn-
hetic temperature profiles computed using the Modular Ocean Data
ssimilation System (MODAS), which models the time-averaged co-
ariability of SSH and subsurface temperature at a given location (Fox
t al., 2002). Salinity is then estimated from the synthetic tempera-
ure profiles using temperature-salinity regression relationships de-
ived from the historical profiles archived in the MODAS database.
Global HYCOM is eddy resolving with an equatorial horizontal
esolution of 0.08° (1/12.5°). There are 32 hybrid vertical coordi-ate layers with potential density referenced to 2000 m, the same as
he present operational US Navy Global Ocean Forecast System ver-
ion 3.0 (Metzger et al., 2014). The surface wind and thermal forcing
re the 0.3125° 1-hourly Climate Forecast System Reanalysis (CFSR)roducts provided by National Centers for Environmental Prediction
NCEP) (Saha et al., 2010). The ocean reanalysis was initialized from
non-assimilative global HYCOM simulation spun-up to statistical
quilibrium using a climatology of NCEP CFSR forcing. The data as-
imilation began on October 1, 1992 and the mesoscale eddy field ad-
usted to the satellite altimeter data within the first month. We ana-
yzed model output over the period January 1993 through December
012.
. Model comparisons with observational data
In this section, we compare HYCOM reanalysis results with ob-
ervational results which were obtained by Na et al. (2014) during
wo years from June 2009 to June 2011 at an array of current and
ressure-recording inverted echo sounders (CPIESs) and CM moor-
ngs. The cross-section, formed with four CPIESs (ES1 to ES4, red dots
n Fig. 1) and three CMs (CM1-3, white dots in Fig. 1), is located be-
ween ES1 and ES4. The HYCOM transect starts from the grid point
earest to ES1 and ends at the grid point closest to ES4, forming a 45°ngle with respect to due east (Fig. 1, gray line). To be consistent with
bservations, a 72-h low-pass filter was applied to the daily transport
ime series from the HYCOM reanalysis.
.1. Current velocity in Kerama Gap
Three CMs mentioned above and deployed during the 2-year ob-
ervational period are CM1 to CM3 (from southwest to northeast)
ith ∼15 km spacing between each mooring. The HYCOM grid pointslosest to the location of the three CMs are chosen to represent the
odel location of the CMs. Velocity time series are extracted from
he “model CM” locations, linearly interpolated to the CM depth, and
hen temporally averaged and compared with observations (Fig. 2).
he average distance between the “model CM” and the deployed CM
ocation is ∼3 km, larger than one third of the grid interval. Given theery steep cross-channel bathymetry and a ∼50 km channel width,t can be difficult to obtain a good point-to-point model-data com-
arison. Below we first compare the yearly averaged currents at each
M in different layers and then provide the correlation coefficient to
etermine the temporal agreements between the reanalysis and ob-
erved current time series. The 2-year observational period is not suf-
ciently long to discuss the annual variation component. So we only
ocus on the mesoscale eddy and the Kuroshio meander components
or the 2-year observational period.
In the upper (Fig. 2a and b) and middle (Fig. 2c and d) layers of
erama Gap, both the observations (black) and reanalysis (red) show
he strongest mean currents at CM3 (the northeasternmost CM). Fol-
owing Na et al. (2014) analysis, we divide the period into year-1 (June
009–June 2010) and year-2 (June 2010–June 2011). In both year-
and year-2, the mean currents gradually decrease from northeast
CM3) toward the southwest (CM1) in Kerama Gap. The magnitude of
he mean currents is reproduced better than the mean current direc-
ion. This discrepancy may be due to the topographic difference be-
ween reality and numerical model, as the current direction is more
ighly sensitive to the local topography compared with the current
peed.
In the upper layer (Fig. 2a and b), the observations in year-1 show
hat mean currents at CM2 flow more normal to rather than parallel
o the mean CM3 current direction. The HYCOM reanalysis correctly
eproduces this characteristic. The year-2 reanalysis accurately repro-
uces the mean current direction in the upper layer for both CM1 and
M3.
In the middle layer (Fig. 2c and d), mean current directions in both
he observations and HYCOM reanalysis are almost parallel to each
ther at CM2 and CM3 while the northwestward mean current direc-
ion in observations is not reproduced in the HYCOM reanalysis. At
M1, the HYCOM reanalysis mean current in year-1 shows weak out-
ow (2.3 cm/s) that is different from the observations showing even
eaker inflow (0.7 cm/s). The HYCOM reanalysis mean current direc-
ion (into the ECS) is more consistent with the observations during
ear-2 than year-1 and the HYCOM reanalysis reproduces observa-
ions that the inflow at CM1 in year-2 is larger than in year-1.
The biggest discrepancy between the HYCOM reanalysis and cur-
ent observations is in the deep layer (Fig. 2e and f), near the bot-
om. Though both the reanalysis and observations show mean inflow
hrough Kerama Gap into ECS, the inflow magnitude (17.3 cm/s) of
he reanalysis is much larger than observed (2.5 cm/s) at the cen-
er of Kerama Gap (CM2). The HYCOM reanalysis appears to have
-
206 Z. Yu et al. / Ocean Modelling 96 (2015) 203–213
Latitude
Dep
th (
m)
(a)
25.7 25.8 25.91500
1000
500
0
Latitude
(b)
25.7 25.8 25.9Latitude
(c)
25.7 25.8 25.9
(m/s)−0.25 −0.2 −0.15 −0.1 −0.05 0 0.05 0.1 0.15 0.2 0.25
Fig. 3. Vertical structure of velocity (m/s) normal to the Kerama Gap transect for (a) the global HYCOM reanalysis, (b) Pacific assimilative HYCOM hindcast, and (c) linear interpo-
lation of the observations (Na et al., 2014). The time frame spans the observational period, June 2009–June 2011.
Table 1
Correlation coefficient between observed and modeled along Kerama Gap ve-
locity at different CMs and layers during June 2009–June 2011 for the two pe-
riod bands: mesoscale eddy (eddy, 70–345 days) and Kuroshio meander (me-
ander, < 70 days) band. The correlation coefficients are all significant to the
95% confidence level.
CM1 CM2 CM3
Eddy Meander Eddy Meander Eddy Meander
Upper 0.66 0.22 0.62 0.29 0.67 0.26
Middle 0.39 0 0.68 0.23 0 0.37
Deep N/A N/A 0 0.19 N/A N/A
o
(
m
p
r
K
p
l
t
d
h
t
3
i
g
t
t
f
t
t
c
w
o
a
G
s
bottom-trapped inflow at CM2 with the maximum occurring near
the sill depth (Fig. 3a) while observations (Fig. 3c) do not show
this feature. The cause of the excessive deep flow appears to be re-
lated to the use of MODAS synthetic profiles within NCODA that are
used for projecting surface information downward into the water
column. Cummings (2005) notes that MODAS has marginal skill in
areas where profiles are limited, and the historical database seems
inadequate to statistically represent the Ryukyu Current in the vicin-
ity of Kerama Gap. A data-assimilative Pacific basin HYCOM hindcast
spanning the Na et al. (2014) observational time period uses an im-
proved methodology, Improved Synthetic Ocean Profiles (ISOP), for
the downward projection of surface information (Helber et al., 2013)
and shows better current structure agreement (Fig. 3b) with observa-
tions than the reanalysis, which supports the above explanation.
Nakamura et al. (2013) compared currents in the deep layer (black
arrow in Fig. 2e and f) with currents observed by ES5 (location shown
in Fig. 1, green dot) at a depth of 1366 m, 50 m above the sea floor, and
suggested a thin vertical layer near the bottom with intensified inflow
across Kerama Gap. Results from the data-assimilative Pacific HYCOM
hindcast (Fig. 3b) agrees with this suggestion, though the currents
in the bottom layer are not as strong as observed and the area with
intensified bottom flow exists only on the northeastern sill.
Thoppil et al. (2015) compared the reanalysis results with 3.5
years of moored CM observations (Ryukyu currents) during Decem-
ber 1998 through October 2002 to the southeast of Amami-Ohshima
Island (Ichikawa et al., 2004). Their comparison has shown good
agreement at depths of 2000, 3000, and even below 4000 m (at these
depths, the flow is not constrained by the data assimilation). Thus the
mismatch of the bottom current through Kerama Gap should not be
interpreted against the deep circulation of the reanalysis in general
but rather confirms that MODAS has marginal skill in areas where
profiles are limited.
The time series of the mesoscale eddy and Kuroshio meander
components are calculated respectively by applying a band pass filter
for periods of 70–345 days and a high pass filter for periods shorter
than 70 days to the time series of total velocity component along
Kerama Gap. The Fourier filter takes the Fourier transform of the time
series, manipulates the specific frequency components, and finally
inverse transforms the results. The correlation coefficients between
reanalysis and observed along Kerama Gap velocity are summarized
in Table 1 for the two different components. The correlation coeffi-
cients are all significant to the 95% confidence level calculated based
n a student t distribution and 50–74 equivalent degrees of freedom
EDOF) for the eddy component and 402-654 EDOF for the Kuroshio
eander component. It can be seen that the mesoscale eddy com-
onents are highly correlated and have a much higher positive cor-
elation coefficient (greater than 0.62 in the upper layer) than the
uroshio meander components (between 0.22 and 0.29 in the up-
er layer) in general except in the deep layer (CM2) and the middle
ayer of CM3. Mesoscale eddies are well integrated into the reanalysis
hrough SSH data assimilation. The less significant correlation in the
eep layer of CM2 and the middle layer of CM3 reflects that MODAS
as marginal skill for projecting surface information downward into
he water column in the northeast Kerama Gap.
.2. Volume transport through Kerama Gap
Volume transport through a zonal (meridional) HYCOM transect
s calculated as the product of the meridional (zonal) depth inte-
rated barotropic velocity and the transect length. Volume transport
hrough a diagonal HYCOM transect is estimated as a sum of the
ransport through the zonal and meridional transects, which starts
rom either end of the diagonal transect and ends at where the two
ransects intersect. Na et al. (2014) estimate that 60% of the mean
ransport is in the upper 500 m. The global HYCOM reanalysis indi-
ates that 61% of the mean transport is in the upper 750 m (Fig. 3a),
hereas the data-assimilative Pacific HYCOM hindcast indicates 65%
f the mean transport is in the upper 500 m (Fig. 3b), a better
greement with the observations. This Pacific hindcast has a Kerama
ap mean transport of 2.05 Sv that also closely agrees with the ob-
ervational estimate. Thus the mean inflow into the ECS does not
-
Z. Yu et al. / Ocean Modelling 96 (2015) 203–213 207
1995 1997 2000 2002 2005 2007 2010 2012−20
−10
0
10
20
Tot
al T
rans
port
(S
v)
Year
Jul09 Oct09 Jan10 Apr10 Jul10 Oct10 Jan11 Apr11−20
−10
0
10
20
Tot
al T
rans
port
(S
v)
Time
CorrCoef = 0.71
mean(yr1)= 1.6 Svmean(yr1)= 1.2 Sv
mean(yr2)= 2.6 Svmean(yr2)= 2.8 Sv
Model
Obs.
a
b
Fig. 4. Comparison of daily mean transport time series from the HYCOM reanalysis (red) with observations (black) from Na et al. (2014) through Kerama Gap. (a) The 20-year time
series from 1993 to 2012, and (b) the 2-year observational period from June 2009 to June 2011. A 72-h low-pass filter has been applied to both time series. Positive transport is flow
into the ECS and negative transport is flow into the Pacific through Kerama Gap. Mean values are listed in their respective colors. (For interpretation of the references to color in
this figure legend, the reader is referred to the web version of this article).
Table 2
Statistics of transport (Sv) through the Kerama Gap. Year-1 is defined as the period from June 2009 to June
2010 and Year-2 is defined as June 2010–June 2011.
2-year mean Year-1 mean Year-2 mean 2 year std Year-1 std Year-2 std
Observation 2.0 1.2 2.8 3.2 2.6 3.6
HYCOM 2.1 1.6 2.6 4.2 3.4 4.8
a
t
b
K
2
s
F
t
(
w
(
r
f
s
c
l
c
T
t
t
r
t
o
3
t
t
(
p
i
4
p
s
w
w
n
a
v
c
r
t
2
c
h
r
n
ppear to be sensitive to the vertical structure of the currents and
he conclusions drawn from the 20-year reanalysis are not impacted
y the difference in the flow structure at the deep layer.
The HYCOM 20-year-long time series of volume transport through
erama Gap is shown in Fig. 4a (red line). The time series during
-year observation period is shown in Fig. 4b. The HYCOM reanaly-
is (red line in Fig. 4b) agrees well with observations (black line in
ig. 4b) but has slightly larger mean and variability. The mean to-
al transport through Kerama Gap in the 2-year hindcast is 2.1 Sv
Table 2) into the ECS while the observation is 2.0 Sv, which is
ell within the standard error of estimate from the observations
0.7 Sv). The estimation error of the 2-year hindcast transport with
espect to observed transport is ±0.4 Sv based on the auto-covarianceunction (Dewar and Bane, 1985) of the transport difference time
eries.
During the 2-year observational period, the HYCOM reanalysis
aptures the temporal transport variability well (Fig. 4b). The corre-
ation coefficient between these two time series is 0.71, and this in-
reases to 0.82 after applying a 20-day smoother to both time series.
he correlation coefficient between the reanalysis and observation
ime series is 0.88 for the mesoscale eddy component, and 0.41 for
he Kuroshio meander component.
Statistics for the two time series are shown in Table 2. The HYCOM
eanalysis mean transports and standard deviations are all higher
han the observed values except the mean transport for the sec-
nd year. The HYCOM 2-year transport standard deviation is 4.2 Sv,
1% greater than the observed value. The HYCOM reanalysis shows
hat there is more flow into the ECS through Kerama Gap and more
ransport variation in year-2 (June 2010–June 2011) than in year-1
June 2009–June 2010), in accord with the observations. The trans-
ort difference between year-1 and year-2 will be discussed further
n Section 5.2.
. Transport variability
The variance preserving power spectra of the 20-year-long trans-
ort time series is shown in Fig. 5. We divide the transport time
eries into four period bands: (1) inter-annual variation component
ith periods longer than 400 days, (2) annual variation component
ith periods between 345 and 400 days, (3) mesoscale eddy compo-
ent with periods between 70 and 345 days, and (4) Kuroshio me-
nder component with periods shorter than 70 days. Most of the
ariation comes from the mesoscale eddy and Kuroshio meander
omponents, which explain 41.3% and 43.9% of the total variance,
espectively. The inter-annual component accounts for 12.5% of the
otal variance, while the annual variation component contains only
.3% of the total variance. In this section, we focus on the seasonal
ycle of volume transport through Kerama Gap (Fig. 6), and examine
ow it is affected by each of the components mentioned above (Fig. 7,
ed, green, and blue lines) except the inter-annual variation compo-
ent (Fig. 7, black dashed line).
-
208 Z. Yu et al. / Ocean Modelling 96 (2015) 203–213
1000 500 200 100 50 20 10 50
0.5
1
1.5
Period (days)
Var
ianc
e (S
v2)
Fig. 5. Variance preserving power spectra (Sv2) of the 20-year global HYCOM reanal-
ysis transport time series through Kerama Gap. Vertical lines show the boundaries of
the temporal bands investigated, i.e. 70, 345, and 400 days.
0
1
2
3
4a
0 2 4 6 8 10 12−1.5
−1
−0.5
0
0.5
1
Vol
ume
Tra
nspo
rt(S
v)
Month
b
Total
Annual
Eddy
Meander
Fig. 6. (a) Seasonal cycle of transport through Kerama Gap and (b) the contribution of
different signals to the transport seasonal cycle from the 1/12.5° global HYCOM reanal-ysis. The shaded area in (a) represents the seasonal cycle uncertainty and the dashed
line shows the 20-year mean transport. (For interpretation of the references to color in
this figure, the reader is referred to the web version of this article).
f
O
(
4
i
t
a
1
T
c
i
r
d
4
2
(
t
t
a
K
7
t
p
(
s
t
J
M
b
i
4
K
m
o
N
f
(
a
o
t
v
t
w
M
5
5
M
b
d
i
(
t
4.1. Mean transport and seasonal cycle
The HYCOM 20-year mean transport through Kerama Gap is
1.95 Sv into the ECS, with a standard deviation of 4.0 Sv. The uncer-
tainty estimation of the 20-year mean transport is ± 0.28 Sv. Thus the1.95 Sv mean is statistically significant.
By averaging the total transport month-by-month over the 20-
year period, we obtain the 20-year mean seasonal cycle of transport
through Kerama Gap (Fig. 6a, black solid line) and the associated
uncertainty (Fig. 6a, shaded area). The most statistically significant
eature of the seasonal cycle is that the maximum transport occurs in
ctober (3.04 Sv) followed by the minimum transport in November
0.54 Sv).
.2. Annual variation component
We obtain the annual variation component (Fig. 7, red) by apply-
ng a band pass Fourier filter for periods of 345-400 days to the total
ransport time series. Explaining only 2.3% of the total variance, the
mplitude of the annual variation component is as large as 2.0 Sv in
997 and as small as 0.3 Sv in 2002. The standard deviation is 0.77 Sv.
he 20-year monthly mean of the annual variation component indi-
ates a clear cycle through Kerama Gap: inflow (0.34 Sv) into the ECS
n summer and outflow (-0.34 Sv) from the ECS in winter (Fig. 6b,
ed line). The total transport inflow (Fig. 6a, black line) in summer is
ominated by the annual variation component (Fig. 6b, red line).
.3. Mesoscale eddy component
Previous studies (Feng et al., 2000; Zhang et al., 2001; Hsin et al.,
008; Lee et al., 2013) have attributed the long-term intra-annual
periods longer than 70 days) Kuroshio transport variability to in-
erior Pacific mesoscale eddies. Similarly, Na et al. (2014) reported
hat the impinging mesoscale eddies from the interior Pacific Ocean
re responsible for the long-term intra-annual variability through
erama Gap. Thus we apply a band pass Fourier filter for periods of
0–345 days to the total transport time series and name the filtered
ime series the mesoscale eddy component (Fig. 7, green). This com-
onent has a standard deviation of 2.90 Sv.
The 20-year monthly mean of the mesoscale eddy component
Fig. 6b, green line) follows the seasonal cycle closely (Fig. 6a, black
olid line). Its contribution to the seasonal cycle can be divided into
hree different stages: (1) neutral stage with a small magnitude from
une through September, (2) an inflow stage in January, February,
ay, and October, and (3) an outflow stage in March, April, Novem-
er, and December. The maximum occurs in October with 1.0 Sv of
nflow and the minimum occurs in November with 0.9 Sv of outflow.
.4. Kuroshio meander component
The short-term intra-annual (periods shorter than 70 days)
uroshio fluctuations have been attributed to two types of Kuroshio
eanders: (1) variations of the Kuroshio path meander with peri-
ds between 30 and 70 days (Ichikawa, 2001; Zhang et al., 2001;
akamura et al., 2003), and (2) variations of the Kuroshio subsur-
ace temperature frontal meander with periods shorter than 30 days
Sugimoto et al., 1988; Qiu et al., 1990; James et al., 1999; Feng et
l., 2000). In this study, we apply a high pass Fourier filter for peri-
ds shorter than 70 days to the total transport time series to obtain
he Kuroshio meander component (Fig. 7, blue line). The standard de-
iation of this component is 2.97 Sv, larger than both annual varia-
ion and mesoscale eddy components. The transport variation related
ith this component (Fig. 6b, blue line) has a smaller magnitude from
ay to September compared to the rest.
. Discussion
.1. Transport through Kerama Gap in relation to transport through
iyakojima to Okinawa and the PN line
Kerama Gap has been suggested as a key region for interaction
etween the ECS-Kuroshio and the Ryukyu Current (Nitani 1972; An-
res et al. 2008a, b; Jin et al. 2010), but the deep gap’s width (∼50 km)s much less than the total distance from Miyakojima to Okinawa
∼250 km). The HYCOM 20-year reanalysis provides an opportunityo compare the mean transport through the smaller passage (Kerama
-
Z. Yu et al. / Ocean Modelling 96 (2015) 203–213 209
1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
−15
−10
−5
0
5
10
15
Vol
ume
Tra
nspo
rt (
Sv)
Year
Inter−annual
Annual
Eddy
Meander
Fig. 7. The transport inter-annual variation component (black dashed line, 20-year mean removed), annual variation component (red), mesoscale eddy component (green), and
Kuroshio meander component (blue) from 1993 to 2012 through Kerama Gap from the 1/12.5° global HYCOM reanalysis. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of this article).
1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 20123
3.5
4
4.5
5
5.5
6
Year
Tra
nspo
rt S
tand
ard
Dev
iatio
n (S
v)
r = 0.64
KGPPN
Fig. 8. Comparison of yearly transport standard deviation between transport through Kerama Gap (solid line) and the PN line (dashed line) for 1993–2012 from the 1/12.5° globalHYCOM reanalysis.
G
t
2
p
t
G
b
K
v
d
e
r
t
c
5
1
C
n
i
t
a
m
a
fi
i
fi
a
s
(
s
a
K
c
G
t
T
i
y
c
f
a
a
t
5
t
ap) and the larger passage (from Miyakojima to Okinawa). Mean
ransport through the passage between Miyakojima and Okinawa is
.03 Sv into ECS, with a standard deviation of 5.74 Sv. The mean trans-
ort through Kerama Gap represents 96% of the mean transport be-
ween Miyakojima and Okinawa. Thus the transport through Kerama
ap is confirmed to be a good approximation to the mean transport
etween Miyakojima and Okinawa, as mentioned earlier in Section 1.
To confirm the conclusions derived from observations that
erama Gap transport may have a significant impact on the temporal
ariability of the Kuroshio transport in the ECS, we calculate the stan-
ard deviation of the transport through Kerama Gap and the PN line
very year. The two time series of standard deviation are highly cor-
elated with a correlation coefficient of 0.64 (Fig. 8), confirming that
he temporal variability of the Kuroshio transport in the ECS (PN line)
orresponds well with the transport variation through Kerama Gap.
.2. Transport in year-1 vs. year-2
In this section, we explain a possible mechanism underlying the
.0 Sv inflow increase from year-1 (1.6 Sv) to year-2 (2.6 Sv) in the HY-
OM reanalysis. The yearly averaged inter-annual variation compo-
ent (black dashed line in Fig. 7 in which the 1.95 Sv mean transport
s removed) is 0.3 Sv in year-2 and -0.3 Sv in year-1, and thus explains
he transport increase of 0.6 Sv from year-1 to year-2. The 2-year aver-
ge inter-annual signal is zero, which helps to explain why the 2-year
ean transport is almost the same as the 20-year mean. Cummings
nd Smedstad (2013) have already verified that the assimilated SSH
eld in the Kuroshio region shows good agreement with independent
nfrared frontal analyses performed by the Naval Oceanographic Of-
ce. Thus we treat the assimilated SSH as the “true” state. The yearly
veraged SSH difference, defined as SSH in year-2 minus in year-1,
hows an anomalous cyclone to the south-southeast of Kerama Gap
Fig. 9) with an SSH difference across the HYCOM Kerama Gap tran-
ect of 1.1 cm. The correlation coefficient between the SSH difference
cross the HYCOM Kerama Gap transect and the transport through
erama Gap is 0.83 over the 20-year reanalysis period. The regression
oefficient between the SSH difference across the HYCOM Kerama
ap transect and transport is 0.46 Sv/cm. Thus the SSH difference be-
ween year-2 and year-1 explains 0.5 Sv of the transport difference.
his indicates that the difference between yearly averaged transport
n year-1 and year-2 corresponds well with the difference between
early averaged SSH differences in year-1 and year-2. It can be con-
luded that about one half of the increase of yearly averaged inflow
rom year-1 to year-2 can be attributed to the increase of yearly aver-
ged inter-annual variation component of inflow transport which is
ccompanied with the development of an anomalous cyclonic eddy
o the south-southeast of Kerama Gap from year-1 to year-2.
.3. Ekman dynamics
The dynamics underlying the annual variation component are at-
ributed to the wind-driven Ekman transport. The mean winds over
-
210 Z. Yu et al. / Ocean Modelling 96 (2015) 203–213
Longitude
Latit
ude
25 cm/s
125 125.5 126 126.5 127 127.5 128 128.5 12924
24.5
25
25.5
26
26.5
27
27.5
28
(cm
)
−8
−4
0
4
8
Fig. 9. SSH difference (year-2–year-1, cm) and layer-1 velocity difference (year-2–year-
1, cm/s) between year-2 (June 2010–June 2011) and year-1 (June 2009–June 2010) from
the global HYCOM reanalysis.
p
d
i
Z
t
s
T
t
n
p
a
y
f
c
t
t
r
n
s
(
a
s
w
(
G
t
D
D
e
i
n
d
o
W
s
t
c
w
a
c
O
w
the broad shelf of the ECS are dominated by the East Asia monsoon. In
summer (June–August), the wind is north–northwestward at Kerama
Gap (Fig. 10a), while in the winter (December–February) the wind
is southwestward and stronger (Fig. 10b). With the prevailing sea-
sonal wind blowing toward the northwest (southwest) persistently
in summer (winter), water piles up on the northeast (northwest)
flank of Kerama Gap due to Ekman transport. Thus the correspond-
ing geostrophic flow is northwestward (southwestward), parallel to
the wind direction and causes the water to flow into (out of) the
ECS through Kerama Gap. The annual variation component of area-
averaged (between 126.0° and 127.5°E, 25.4° and 26.7°N) monthlyalong Kerama Gap wind stress anomaly (30° counter-clockwise fromnorth) shows good agreement with the monthly transport of the an-
nual variation component (Fig. 10c) and the correlation coefficient
is 0.55. Hsin et al. (2010) found a similar relationship near southeast-
ern Taiwan, where the geostrophic velocity and local meridional wind
stress are generally well-correlated on the seasonal time scale.
5.4. Monthly mean SSH anomaly
Unlike Na et al. (2014), the mesoscale eddy component is rep-
resented by a group of peaks with similar amplitude (Fig. 5) in the
Latit
ude
(oN
)
Longitude (oE)
.1 N/m2
126 127 128
25
25.5
26
26.5
27
27.5
.1 N/m2
Longitude (oE)
126 127 128
ba
Fig. 10. Mean CFSR wind stress vectors averaged over the period 1993–2012 in the ECS du
model land points. The reference vector is shown in the upper left corner of each panel. (c) S
anomaly (N/m2) and monthly transport (Sv) of the annual variation component through Kera
eriod band 70–200 days, instead of a single dominant peak at ∼100-ay period. The period band 70–200 days agrees well with the dom-
nant time scale of the transport mode through the ETC reported by
hang et al. (2001). Spectral analysis using the reanalysis transport
ime series from only the 2-year observational period does show a
ingle dominant peak at ∼100-day period, the same as observations.his indicates that time intervals of arriving mesoscale eddies from
he interior ocean vary with time over the 20-year period, and are
ot dominated by a single 100-day time period.
In order to explain the monthly mean of the mesoscale eddy com-
onent (green line in Fig. 6b), we generated a monthly mean SSH
nomaly (SSHA) map centered on Kerama Gap (Fig. 11). An EOF anal-
sis was performed and the leading mode of the annual steric ef-
ect (Stammer, 1997) was also removed. Depending on the eddy lo-
ation, the same type of eddy can increase or decrease transport
hrough Kerama Gap. The mesoscale eddies typically propagate into
his region as part of the return flow of the Kuroshio’s non-linear
ecirculation gyre. Therefore, it can be concluded that the Kuroshio’s
on-linear recirculation gyre is one of the possible reasons for the
ignificant month to month variations shown in Fig. 11. Andres et al.
2008a) show that positive transport anomalies in Kerama Gap are
ssociated with the arrival of anticyclonic eddies along the eastern
ide of Okinawa, while negative transport anomalies are associated
ith the arrival of cyclonic eddies. Na et al. (2014) find that cyclonic
anticyclonic) eddies increase (decrease) transport through Kerama
ap when these eddies are located to the south of Kerama Gap.
Below, we examine the reason why the contributions of eddies to
he transport seasonal cycle are large in January–May and October–
ecember (Fig. 6b, green line), and small from June to September.
uring the inflow stage, an anomalous anticyclone is located to the
ast of Okinawa in January and to the southeast of Kerama Gap
n February. A dipole with an anticyclone (cyclone) attached to the
orthern (southern) Kerama Gap is shown in October. When both ed-
ies pump water into ECS through Kerama Gap, the maximum inflow
ccurs in October. But the SSHA map in May shows an exception.
hen a cyclone is located to the east of Okinawa in May, the eddy
hould decrease transport (Andres et al., 2008a) instead of increasing
ransport (Fig. 6b, green line).
A cyclonic eddy located to the east of Okinawa and an anti-
yclonic eddy to the south/southeast of Kerama Gap is consistent
ith outflow from the ECS through Kerama Gap, as occurs in March
nd April. But during November and December, an elongated anti-
yclonic eddy that straddles the entire passage from Miyakojima to
kinawa separates two cyclones located to the southeast and north-
est of Kerama Gap. The anticyclone straddling Kerama Gap would
129
Dep
th (
m)
0
300
600
900
3000
6000
−0.1 −0.05 0 0.05 0.1
−4
−2
0
2
4
Mon
thly
Tra
nspo
rt (
Sv)
Along Channel Wind Stress Anomaly (N/m2)
c
ring summer (a, June–August) and winter (b, December–February). Gray represents
catter plot of annual variation component of monthly along Kerama Gap wind stress
ma Gap.
-
Z. Yu et al. / Ocean Modelling 96 (2015) 203–213 211
Jan
Latit
ude
(oN
)
25
26
27
28Feb Mar
Apr
Latit
ude
(oN
)
25
26
27
28May Jun
Jul
Latit
ude
(oN
)
25
26
27
28Aug Sep
Oct
Longitude (oE)
Latit
ude
(oN
)
125 126 127 128
25
26
27
28Nov
Longitude (oE)125 126 127 128
Dec
Longitude (oE)125 126 127 128
SSH Anomaly (cm)−8 −4 0 4 8
Fig. 11. Monthly mean (1993–2012) SSHA (cm, relative to annual mean SSH with an-
nual steric effect removed) pattern from January to December. Contours with SSHA =2 cm (solid white line) are used to identify the anomalous anticyclones and contours
with SSHA = −2 cm (dashed white lines) are used to identify the anomalous cyclones.
s
m
(
fl
a
t
p
w
G
(
2
a
l
t
d
J
p
t
a
K
t
fl
5
K
N
t
m
c
p
i
g
d
s
p
i
i
2
n
b
2
w
n
p
s
c
F
u
uggest weak flow instead of the strong outflow shown by the
onthly mean.
The vertical structure of the normal velocity anomaly in May
November and December) shows that subsurface water primarily
ows into (out of) the ECS through Kerama Gap. Jin et al. (2010)
10 cm/s
Longitude (oE)
Latit
ude
(oN
)
a
125 126 127 128
24.5
25
25.5
26
26.5
27
27.5
28
10 cm
b
ig. 12. Monthly velocity anomaly (cm/s, relative to annual mean velocity) at 150 m overlaid
pper left corner of each panel.
pplied the self-organizing map to study the interaction between
he ECS and the Ryukyu Current through Kerama Gap. Four coherent
atterns were extracted to illustrate how eddies in the ECS interact
ith eddies in the Ryukyu Current to alter the flow through Kerama
ap. The velocity anomaly structure in May (Fig. 12a) and November
Fig. 12b) belongs to patterns P4 and P3 (Fig. 2d and c in Jin et al.,
010, respectively). Thus the transport anomaly in May, November,
nd December (velocity anomaly structure in December is very simi-
ar to November and is not shown) is caused by eddy interactions on
he western and eastern sides of Kerama Gap and is associated with
eeper levels, but is not represented by the SSHA.
Small eddy contributions to the transport seasonal cycle from
une to September indicate that the impact of eddies on the trans-
ort through Kerama Gap is small during these four months and
he SSHA maps confirm this. The cyclonic eddies are either far
way from Kerama Gap (June and September) or oriented along the
erama Gap transect and thus generate small SSH difference across
he transect (July and August) (and additionally have negligible deep
ow).
.5. Baroclinic instability
Previous studies have shown that baroclinic instabilities lead to
uroshio meander variations (James et al., 1999; Zhang et al., 2001;
akamura et al., 2003; Hsin et al., 2008). Charney (1947) developed
he baroclinic instability theory for large scale quasi-geostrophic at-
ospheric waves while Orlanski and Cox (1973) show us how baro-
linic instability develops when horizontal density gradients are
resent in the ocean. The horizontal density/temperature gradient
s essential since it provides the available potential energy for the
rowth of the baroclinic instability. The horizontal temperature gra-
ient between Kuroshio water and the ambient ECS water shows a
easonal cycle: weak in summer and strong during the winter-spring
eriod (Nagata and Takeshita, 1985). Thus, the baroclinic instability
s suppressed (enhanced) in summer (winter and spring) as shown
n Fig. 6b (blue line). Previous observations (Nakamura et al., 2006,
008) have indicated that the Kuroshio pathway in the northern Oki-
awa Trough is destabilized during the winter–spring period and sta-
ilized during the summer–autumn period. Nakamura et al. (2010,
012) additionally examined this seasonality of the Kuroshio path-
ay destabilization and found that baroclinic instability triggered by
onlinear Ekman divergence due to wind stress in autumn and winter
lays an important role in Kuroshio pathway variation. Thus the ob-
erved transport variability seems to be explained by the theoretical
onsiderations on the internal baroclinic instability.
/s
Longitude (oE)126 127 128
SS
H A
nom
aly
(cm
)
−10
−8
−6
−4
−2
0
2
4
6
8
10
on top of the SSHA in May (a) and November (b). The reference vector is shown in the
-
212 Z. Yu et al. / Ocean Modelling 96 (2015) 203–213
M
p
R
A
A
A
B
C
C
C
C
C
D
F
F
G
G
H
H
H
I
I
J
J
J
K
L
M
M
N
N
N
6. Conclusions
A global HYCOM data-assimilative reanalysis was integrated for
20 years from 1993 to 2012 and used to study the transport variability
through Kerama Gap, the deepest channel in the Ryukyu Islands Arc,
and an important passage for water exchange between the ECS and
the Northwest Pacific. The reanalysis volume transport time series
through Kerama Gap was confirmed to accurately reproduce the 2-
year observational time series from June 2009 to June 2011 reported
by Na et al. (2014). The discrepancy of the bottom velocity between
the reanalysis and observations confirms that MODAS has marginal
skill in areas where profiles are limited. From the 20-year transport
time series of volume transport, we estimated the 20-year monthly
mean seasonal cycle that has the maximum in October (3.0 Sv) and
the minimum in November (0.5 Sv).
The transport time series has large variability with a maximum
of 17.3 Sv (May 1994) and minimum of −14.5 Sv (December 1996).The 20-year mean of the volume transport is 1.95 Sv into the ECS. Its
standard deviation is 4.0 Sv, equal to the observed standard deviation
of the ECS Kuroshio volume transport at the PN-line (Andres et al.,
2008b), which indicates a significant impact of Kerama Gap transport
on the temporal variability of the Kuroshio transport in the ECS.
The annual variation component, with periods between 345 and
400 days, explains only 2.3% of the total transport variance, but it
makes a significant contribution to the seasonal cycle (Fig. 6b, red
line). This variation component tends to accompany an increase of
inflow through Kerama Gap in summer and a decrease in winter. It
is explained by the Ekman dynamics responding to seasonal changes
of the local winds, which contribute a positive transport anomaly in
summer and a negative transport anomaly in winter.
The mesoscale eddy component, with periods between 70 and
345 days, makes the most significant contribution to the trans-
port seasonal cycle except during summer from June to September
(Fig. 6b, green line). The impinging mesoscale eddies substantially
affect the monthly mean, increase the transport into the ECS during
January, February, May, and October, and decrease it in March, April,
November, and December. In summer, contributions of impinging cy-
clonic and anticyclonic eddies are nearly equal to each other, and the
apparent influence of eddies diminishes.
The contribution of the Kuroshio meander components with peri-
ods shorter than 70 days to the seasonal cycle is larger in winter than
in summer. Baroclinic instability was suggested to be one possible
explanation.
The contribution of the inter-annual component to the seasonal
cycle is just the 20-year mean transport (Fig. 6a, black dashed line)
due to its long time period and thus is not discussed in this paper.
However, we will present the inter-annual component of the vol-
ume transport in a separate paper that examines its impact on ex-
treme flow events, i.e., when transport anomaly exceeds one standard
deviation.
Acknowledgments
The numerical output used for this paper can be found on the
http://www.hycom.org data server under the “HYCOM + NCODAGlobal 1/12° Reanalysis” link. This effort was funded by the “6.1Kuroshio and Ryukyu Current Dynamics” project sponsored by the
Office of Naval Research under program element 0601135 N. Z. Y. was
supported by a Post-Doctoral Fellowship from the American Society
for Engineering Education, with funding provided by the Naval Re-
search Laboratory, Stennis Space Center, MS. Computer time was pro-
vided by the Department of Defense (DoD) High Performance Com-
puting Modernization Program and the simulations were performed
on the IBM Power 6 (daVinci) and the IBM iDataPlex (Kilrain) at the
Navy DoD Supercomputing Resources Center, Stennis Space Center,
S. This is NRL contribution NRL/JA/7320-15-2704. It has been ap-
roved for public release and distribution is unlimited.
eferences
ndres, M., Park, J.-H., Wimbush, M., Zhu, X.-H., Chang, K.-I., Ichikawa, H., 2008a. Study
of the Kuroshio/Ryukyu current system based on satellite-altimeter and in situmeasurements. J. Oceanogr. 64, 937–950.
ndres, M., Wimbush, M., Park, J.-H., Chang, K.-I., Lim, B.-H., Watts, D.R., Ichikawa, H.,Teague, W.J., 2008b. Observations of Kuroshio flow variations in the East China Sea.
J. Geophys. Res. 113, C05013. doi:10.1029/2007JC004200.
ndres, M., Cenedese, C., 2013. Laboratory experiments and observations of cyclonicand anticyclonic eddies impinging on an island. J. Geophys. Res. Oceans 118. doi:10.
1002/jgrc.20081.leck, R., 2002. An oceanic general circulation model framed in hybrid isopycnic-
Cartesian coordinates. Ocean Modell. 4, 55–88.harney, J.G., 1947. The dynamics of long waves in a baroclinic westerly current. J. Me-
teorol. 4, 135–162.
hassignet, E.P., Smith, L.T., Halliwell, G.R., Bleck, R., 2003. North Atlantic simulationswith the Hybrid Coordinate Ocean Model (HYCOM): impact of the vertical coordi-
nate choice, reference pressure, and thermobaricity. J. Phys. Oceanogr. 33, 2504–2526.
hoi, B.H., Kim, K.O., Eum, H.M., 2002. Digital bathymetric and topographic data forneighboring seas of Korea. J. Korean Soc. Coast. Ocean Eng. 14, 41–50.
ummings, J.A., 2005. Operational multivariate ocean data assimilation. Q. J. R. Meteo-
rol. Soc. 131, 3583–3604.ummings, J.A., Smedstad, O.M., 2013. Variational data assimilation for the global
ocean. In: Park, S.K., Xu, L. (Eds.). Data Assimilation for Atmospheric, Oceanic,and Hydrologic Applications (Vol. II). Springer-Verlag, Berlin, Heidelberg http:
//dx.doi.org/10.1007/978-3-642-35088-7_13 .ewar, W.K., Bane, J.M., 1985. Subsurface energetics of the Gulf Stream near the
Charleston Bump. J. Phys. Oceanogr. 15, 1771–1789.eng, M., Mitsudera, H., Yoshikawa, Y., 2000. Structure and variability of the Kuroshio
current in Tokara Strait. J. Phys. Oceanogr. 30, 2257–2276.
ox, D.N., Teague, W.J., Barron, C.N., Carnes, M.R., Lee, C.M., 2002. The modular oceandata assimilation system. J. Atmos. Ocean. Technol. 19, 240–252.
ordon, A.L., Flament, P., Villanoy, C., Centurioni, L., 2014. The nascent Kuroshio of La-mon Bay. J. Geophys. Res. Oceans 119, 4251–4263. doi:10.1002/2014JC009882.
uo, X., Miyazawa, Y., Yamagata, T., 2006. The Kuroshio onshore intrusion along theshelf break of the East China Sea: the origin of the Tsushima warm current. J. Phys.
Oceanogr. 36, 2205–2231.
elber, R.W., T.L. Townsend, C.N. Barron, J.M. Dastugue, and M.R. Carnes, 2013.Validation Test Report for the Improved Synthetic Ocean Profile (ISOP)
System: Part I. Synthetic Profile Methods and Algorithm. NRL memoran-dum report NRL/MR/7320—13-9364, http://www7320.nrlssc.navy.mil/pubs/2013/
helber1-2013.pdf.sin, Y.-C., Wu, C.-R., Shaw, P.-T., 2008. Spatial and temporal variations of the Kuroshio
east of Taiwan, 1982–2005: a numerical study. J. Geophys. Res. 113, C04002. doi:10.
1029/2007JC004485.sin, Y.-C., Qu, T., Wu, C.-R., 2010. Intra-seasonal variation of the Kuroshio southeast of
Taiwan and its possible forcing mechanism. Ocean Dyn. 60, 1293–1306.chikawa, H., Nakamura, H., Nishina, A., Higashi, M., 2004. Variability of north-eastward
current southeast of northern Ryukyu Islands. J. Oceanogr. 60, 351–363.chikawa, K., 2001. Variation of the Kuroshio in the Tokara Strait induced by meso-scale
eddies. J. Oceanogr. 57, 55–68.
ames, C., Wimbush, M., Ichikawa, H., 1999. Kuroshio meanders in the East China Sea.J. Phys. Oceanogr. 29, 259–272.
in, B., Wang, G., Liu, Y., Zhang, R., 2010. Interaction between the East China SeaKuroshio and the Ryukyu Current as revealed by the self-organizing map. J. Geo-
phys. Res. 115, C12047. doi:10.1029/2010JC006437.ohns, W.E., Lee, T.N., Zhang, D., Zantopp, R., Liu, C.-T., Yang, Y., 2001. The Kuroshio east
of Taiwan: moored transport observations from the WOCE PCM-1 array. J. Phys.
Oceanogr. 31, 1031–1053.undu, P.K., Cohen, I.M., 2002. Fluid Mechanics. Academic Press, San Diego.
ee, I.-H., Ko, D.S., Wang, Y.-H., Centurioni, L., Wang, D.-P., 2013. The mesoscale eddiesand Kuroshio transport in the western North Pacific east of Taiwan from 8-year
(2003–2010) model reanalysis. Ocean Dyn. 63, 1027–1040.etzger, E.J., Smedstad, O.M., Thoppil, P.G., Hurlburt, H.E., Cummings, J.A.,
Wallcraft, A.J., Zamudio, L., Franklin, D.S., Posey, P.G., Phelps, M.W., Hogan, P.J.,
Bub, F.L., Dehaan, C.J., 2014. US Navy operational global Ocean and Arctic ice pre-diction systems. Oceanography 27, 32–43.
orinaga, K., Nakagawa, N., Osamu, K., Guo, B., 1998. Flow pattern of the Kuroshiowest of the main Okinawa Island. In: Proceedings of Japan–China Joint Symposium
on Cooperative Study of Subtropical Circulation System. Seikai National FisheriesResearch Institute, Nagasaki, Japan, pp. 203–210.
a, H., Wimbush, M., Park, J.-H., Nakamura, H., Nishina, A., 2014. Observations of flowvariability through the Kerama Gap between the East China Sea and the north-
western Pacific. J. Geophys. Res., Oceans 119, 689–703. doi:10.1002/2013JC008899.
agata, U., Takeshita, K., 1985. Variation of the sea surface temperature distributionacross the Kuroshio in the Tokara Strait. J. Oceanogr. Soc. Jpn. 41, 244–258.
akamura, H., Ichikawa, H., Nishina, A., Lie, H.-J., 2003. Kuroshio path meander be-tween the continental slope and the Tokara Strait in the East China Sea. J. Geophys.
Res. 108 (C11), 3360. doi:10.1029/2002JC001450.
http://www.hycom.orghttp://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0001http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0001http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0001http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0001http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0001http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0001http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0001http://dx.doi.org/10.1029/2007JC004200http://dx.doi.org/10.1002/jgrc.20081http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0004http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0004http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0005http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0005http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0006http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0006http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0006http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0006http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0006http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0007http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0007http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0007http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0007http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0008http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0008http://dx.doi.org/10.1007/978-3-642-35088-7_13http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0010http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0010http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0010http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0011http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0011http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0011http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0011http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0012http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0012http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0012http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0012http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0012http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0012http://dx.doi.org/10.1002/2014JC009882http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0014http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0014http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0014http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0014http://www7320.nrlssc.navy.mil/pubs/2013/helber1-2013.pdfhttp://dx.doi.org/10.1029/2007JC004485http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0016http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0016http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0016http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0016http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0017http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0017http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0017http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0017http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0017http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0018http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0018http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0019http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0019http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0019http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0019http://dx.doi.org/10.1029/2010JC006437http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0021http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0021http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0021http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0021http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0021http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0021http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0021http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0022http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0022http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0022http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0023http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0023http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0023http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0023http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0023http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0023http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0024http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0025http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0025http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0025http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0025http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0025http://dx.doi.org/10.1002/2013JC008899http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0027http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0027http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0027http://dx.doi.org/10.1029/2002JC001450
-
Z. Yu et al. / Ocean Modelling 96 (2015) 203–213 213
N
N
N
N
N
N
O
O
Q
Q
Q
R
S
S
S
S
T
Y
Y
Y
Z
akamura, H., Nishina, A., Ichikawa, H., 2006. Time-frequency variability ofKuroshio meanders in Tokara Strait. Geophys. Res. Lett. 33, L21605. doi:10.1029/
2006GL027516.akamura, H., Nishina, A., Ichikawa, H., Nonaka, M., Sasaki, H, 2008. Deep countercur-
rent beneath the Kuroshio in the Okinawa trough. J. Geophys. Res. 113, C06030.doi:10.1029/2007JC004574.
akamura, H., Nonaka, M., Sasaki, H., 2010. Seasonality of the Kuroshio path destabi-lization phenomenon in the Okinawa Trough: a numerical study of its mechanism.
J. Phys. Oceanogr. 40, 530–550. doi:10.1175/2009JPO4156.1.
akamura, H., Nishina, A., Tabata, K., Higashi, M., Habano, A., Yamashiro, T., 2012.Surface velocity time series derived from satellite altimetry data in a section
across the Kuroshio southwest of Kyushu. J. Oceanogr. 68, 321–336. doi:10.1007/s10872-012-0101-4.
akamura, H., Nishina, A., Liu, Z., Tanaka, F., Wimbush, M., Park, J.-H., 2013. Intermedi-ate and deep water formation in the Okinawa Trough. J. Geophys. Res. 118, 6881–
6893.
itani, H., 1972. Beginning of the Kuroshio. Kuroshio. University of Washington Press,Seattle, pp. 129–163.
ka, E., Kawabe, M., 1998. Characteristics of variations of water properties and densitystructure around the Kuroshio in the East China Sea. J. Oceanogr. 54, 605–617.
rlanski, I., Cox, M., 1973. Baroclinic instability in ocean currents. Geophys. Fluid Dyn.4, 297–332.
iu, B., 2001. Kuroshio and Oyashio Currents. Encyclopedia of Ocean Sciences, 3. Aca-
demic Press, San Diego, CA, pp. 1413–1425.iu, B., Toda, T., Imasato, N., 1990. On Kuroshio front fluctuations in the East China Sea
using satellite and in situ observational data. J. Geophys. Res. 95, 18191–18204.u, T., Lukas, R., 2003. The bifurcation of the North Equatorial Current in the Pacific. J.
Phys. Oceanogr. 33, 5–18.
oemmich, D., McCallister, T., 1989. Large scale circulation of the North Pacific Ocean.Prog. Oceanogr. 22, 171–204.
aha, S., others, 2010. The NCEP climate forecast system reanalysis. Bull. Am. Meteorol.Soc. 91, 1015–1057 http://dx.doi.org/10.1175/2010BAMS3001.1 .
oeyanto, E., Guo, X., Ono, J., Miyazawa, Y., 2014. Interannual variations of Kuroshiotransport in the East China Sea and its relation to the Pacific decadal oscilla-
tion and mesoscale eddies. J. Geophys. Res. Oceans 119, 3595–3616. doi:10.1002/2013JC009529.
ugimoto, T., Kimura, S., Miyaji, K., 1988. Meander of the Kuroshio front and current
variability in the East China Sea. J. Oceanogr. Soc. Jpn. 44, 125–135.tammer, D., 1997. Steric and wind-induced changes in TOPEX/POSEDON large-scale
sea surface topography observations. J. Geophys. Res. 102, 20,987–21,009.hoppil, P., Metzger, E.J., Hurlburt, H.E., Smedstad, O.M., Ichikawa, H., 2015. The cur-
rent system east of the Ryukyu Islands as revealed by a global ocean reanalysis.Submitted to Prog. Oceanogr 751–753.
ou, S.-H., Yoon, J.-H., 2004. Modeling of the Ryukyu current along the Pacific side of
the Ryukyu Islands. Pacific Oceanogr. 2, 44–51.uan, Y., Takano, K., Pan, Z., Su, J., Kawatate, K., Imawaki, S., Yu, H., Chen, H., Ichikawa, H.,
Umatani, S., 1994. The Kuroshio in the East China Sea and the currents east of theRyukyu Islands during autumn 1991. La Mer 32, 235–244.
uan, Y., Su, J., Pan, Z., Chen, H., Ichikawa, H., Imawaki, S., Kawatate, K., Takano, K.,Umatani, S.-I., 1995. The western boundary current east of the Ryukyu Islands. La
Mer 33, 1–11.
hang, D., Johns, W.E., Lee, T.N., Liu, C.-T., Zantopp, R., 2001. The Kuroshio east of Tai-wan: modes of variability and relationship to interior mesoscale eddies. J. Phys.
Oceanogr. 31, 1054–1074.
http://dx.doi.org/10.1029/2006GL027516http://dx.doi.org/10.1029/2007JC004574http://dx.doi.org/10.1175/2009JPO4156.1http://dx.doi.org/10.1007/s10872-012-0101-4http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0033http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0033http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0033http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0033http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0033http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0033http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0033http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0034http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0034http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0035http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0035http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0035http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0036http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0036http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0036http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0037http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0037http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0038http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0038http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0038http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0038http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0039http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0039http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0039http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0040http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0040http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0040http://dx.doi.org/10.1175/2010BAMS3001.1http://dx.doi.org/10.1002/2013JC009529http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0043http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0043http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0043http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0043http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0044http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0044http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0045http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0045http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0045http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0045http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0045http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0045http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0046http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0046http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0046http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0047http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0048http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0049http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0049http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0049http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0049http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0049http://refhub.elsevier.com/S1463-5003(15)00215-2/sbref0049
Seasonal cycle of volume transport through Kerama Gap revealed by a 20-year global HYbrid Coordinate Ocean Model reanalysis1 Introduction2 Numerical model3 Model comparisons with observational data3.1 Current velocity in Kerama Gap3.2 Volume transport through Kerama Gap
4 Transport variability4.1 Mean transport and seasonal cycle4.2 Annual variation component4.3 Mesoscale eddy component4.4 Kuroshio meander component
5 Discussion5.1 Transport through Kerama Gap in relation to transport through Miyakojima to Okinawa and the PN line5.2 Transport in year-1vs. year-25.3 Ekman dynamics5.4 Monthly mean SSH anomaly5.5 Baroclinic instability
6 Conclusions Acknowledgments References